Topic: [ANN] Bitfury is looking for alpha-testers of first chips! FREE MONEY HERE! - page 44. (Read 176729 times)

Amazing Bitfury!

Congratulations and thanks for holding your word!

Much respect.

It would make logical sense, if it also made technical sense, that the power circuitry allowed for a 'high speed/turbo' mode that would allow to seize on the opportunity of relatively low difficulty at the present moment.


And, given bitfury's forward thinking electrical usage design, the ability for manual user clocking / undervolting to the 6/2 ratio would be equally as beneficial once Bitcoin reaches near Petahash capacity.


I'd be somewhat disappointed if bitfury did not provide circuitry that allowed for both a mutually exclusive hi speed-hi consumption setting and then the 6/2 setting.


My 2 cents.  If the bitfury units possessed both capabilities I'd be more likely to purchase.

Congratulations on a an excellent chip. Now, the only relevant question is when will I be able to purchase one?

These chips are completely irrelevant if restricted to 100TH and Metabank (who wont ship outside of Russia). Either sell us the chips or sell us the devices that use these chips, but stop waving them in our faces and telling us we can't have them.

Very nice, however, I'm also thinking BFLs chips were looking this good this early, they didn't know they had power issues until they tried to run them at a decent frequency.

The question is, where is your "ski jump" where the power/frequency goes exponential? BFL found their's was not as far to the right of the graph as they hoped.

If the "ski jump" is to the right of your intended frequency spec, all well and good, great job. If it's to the left.... well, there's always 14nm.....

An RF digital SHA256 chip?

to intron
https://bitcointalksearch.org/topic/m.2508369
despite the fact that it is the Russian language, I have no idea what is 'BC817-40' or 'level shifter ', so maybe you'll understand more in this post than I 

Hi bitfury,

Saw in your code you are clocking the SPI chain at 200 kHz:

  speed = 200000;

  if (ioctl(fd, SPI_IOC_WR_MAX_SPEED_HZ, &speed) < 0) { perror("Unable to set WR_MAX_SPEED_HZ"); close(fd); return -1; }
  if (ioctl(fd, SPI_IOC_RD_MAX_SPEED_HZ, &speed) < 0) { perror("Unable to set RD_MAX_SPEED_HZ"); close(fd); return -1; }

Why it is so slow? Just te be sure during development or are there other reasons?
It's not a problem, just curious.

intron

OK, I'll try...

Bitfury did just a short burst tests (lasting at most few seconds) of his chip at various voltages and various speeds. He still need to test it under a continuous load and at various temperatures.

Now I switch to explaining what he didn't do. He didn't use any "standard cell", "library" or other purely digital logic EDA tools, which were thus far used by all other competitors. They used unrolled pipeline designs and their simulations were very inaccurate-purely digital. Due to the very approximate simulation and high toggle rate of the flip-flops in SHA-256 the real power used by competitor chips was significantly higher than the simulated power usage.

Now I switch back to what he did do. After completing the initial digital logic design he had switched to a full-custom design process: he no longer designed at the gate or flip-flop level but at the individual transistor level. He no longer dealt with zeros and ones but with volts and amperes. He had used a very precise analog models of the transistors from the foundry and used analog circuit simulators: SPICE and/or BSIM. This was possible without multimillion budget because he didn't unroll the hashing kernels but kept them rolled and replicated them several hundred times over the area of the chip. The additional benefit was that his power simulations were very accurate: the simulation error margins were lower than the wafer manufacture process error margins.

One of the goals he had choosen for his chip was to have them work with rather wide ranges of the supply voltages, in particular with uncommonly low voltages. To achieve that he had paid extreme attention to minimizing the switching noise both inside the chip and on the chip pins. In particular despite of the relatively low clock speed (100MHz or low hundreds MHz) the I/O design methodology was the one used for GHz signals (or high hundreds of MHz).

Because of the above goals his chip isn't really plug-and-play for designers used to deal with typical digital logic design workflows and tools. To get all thats possible out of his chip requires familiarity with analog and mixed signal design. In particular it requires familiarity with transmission line modeling to design the best PCB with a large quantity of his chips. His goal is to publish http://en.wikipedia.org/wiki/Scattering_parameters required to accurately model the I/O terminals of his chip.

He probably already has a good idea of what those S-parameters are from the high accuracy simulations using SPICE/BSIM. But possibly he either can't (because of NDA with the foundry) or doesn't want (for competitive advantage reasons) release his internal models and their parameters. Additionally, the S-parameters are affected by the quality of packaging of the chip and there is no real substitute for making the actual measurements, for which he plans to use devices made by http://www.signalhound.com/ .

Obviously, I'm not bitfury, and the above is just my educated guess.

The translation: Awesome results - just as he intended.

Great respect.

Can someone translate what bitfury said in his long post? As i understood there are more measurements to be done, but the chip is working, but it isn't good enough for a big rig? I can't wait for USB miners with this chip.

Also i'm impressed too for paying for that bet! Now just give me the change to be your customer bitfury!

You were right bitfury, the initial input level shifter circuit
was not working correctly. Seems the the parasitic capacitance
of the 1V8 zener together with the 1K input resistance caused
severe detoriation of the waveform due to the RC-delay.

Redid the circuit: increased the current by lowering the
resistance to ground and replaced the zener with a 3V3
type connected directly to in the inputs. Did also some
measurements, looked not that bad. Inputs SPI signals are
divided to 1V8, output SPI signal is level shifted to 3V3.
See image:



This is still with a bare board, no ASIC yet. Might need
to lower the restistors even further to get crisp edges.

Will first look into the SPI code now.

intron

Bitfury, you are among the very rare honest ASIC sellers with great integrity. Hats off to you, comrade
If I could, I would invest the amount from the bet in even more of your miners.

Great job bitfury chip performs really well do to your measurments.

I've been wracking my brain all night, trying to figure out why I was getting sent 0.1btc from bitbet, then was linked to this thread from the bitcoin subreddit, relating to the testing being done. I have to say I'm impressed with the current results, and have high hopes for this project to continue to develop.

Congratulation!
As a local patriot I'm very impressive that such a good made chip is a child of Slavic country cooperation

Congratulations! I now look forward to the next announcement with you guys selling your chips in bulk so local builders can manufacture goods based around your design. Truly impressive results that bode well for some energy efficient builds.

edit: Someone using your chip is going to kick ASICMINER's USB Eruptor's butt 

best warm wishes to you 

i feel we are all now as Year 1897 when E Reserford and JJ Thomson began nuclear era

Amazing results, congratulations!


Does "100 Th/s operations, etc" include Metabank too?

.b

Congrats bitfury on getting the chip to work!

+1 impressive
 
It is too bad you don't sell your products to the rest of the world though... you allow lucky Russian citizens to charge huge mark ups for customers wanting to buy your product in the rest of the world. If they are going to ship worldwide for a profit, why not charge more and ship worldwide yourselves?

I would have bought some units, but honestly I didn't want to order through some sketchy Jr. member. I trust you Bitfury, but not some random person that is reselling your units for profit...